A critical role of action-related functional networks in Gilles de la Tourette syndrome

Abstract

Gilles de la Tourette Syndrome (GTS) is a chronic tic disorder, characterized by unwanted motor actions and vocalizations. While brain stimulation techniques show promise in reducing tic severity, optimal target networks are not well-defined. Here, we leverage datasets from two independent deep brain stimulation (DBS) cohorts and a cohort of tic-inducing lesions to infer critical networks for treatment and occurrence of tics by mapping stimulation sites and lesions to a functional connectome derived from 1,000 healthy participants. We find that greater tic reduction is linked to higher connectivity of DBS sites (N = 37) with action-related functional resting-state networks, i.e., the cingulo-opercular (r = 0.62; p < 0.001) and somato-cognitive action networks (r = 0.47; p = 0.002). Regions of the cingulo-opercular network best match the optimal connectivity profiles of thalamic DBS. We replicate the significance of targeting cingulo-opercular and somato-cognitive action network connectivity in an independent DBS cohort (N = 10). Finally, we demonstrate that tic-inducing brain lesions (N = 22) exhibit similar connectivity to these networks. Collectively, these results suggest a critical role for these action-related networks in the pathophysiology and treatment of GTS.

Fig. 1. Method Overview.

a As regions of interest, we chose functional networks based on resting-state connectivity known to be involved during actions in humans. The cingulo-opercular network (CON), also referred to as the action-mode network, is linked to the processing of arousal, error detection and pain sensation and thus states that call for action. The somato-cognitive action network (SCAN) has recently been described as a network relevant to action planning and complex body movements. Both networks show high connectivity with each other. b For each subject in a cohort of patients with GTS undergoing thalamic DBS for tic reduction (N = 37), electrodes and the volume of activated tissue (i.e., stimulation site) were reconstructed in standard space. Using a publicly available resting-state functional connectome acquired in healthy participants (N = 1000), we computed the functional connectivity of each bilateral pair of stimulation sites with all other brain voxels. The resulting connectivity maps were then used for group analysis. c Initially (1), a map of voxel-wise average connectivity was computed (termed average-map). Subsequently (2), the average connectivity in priori-defined regions of interest, i.e., CON and SCAN, was correlated with tic reduction post-DBS. In an alternative data-driven whole-brain approach (3), we calculated a map of voxel-wise significant correlations (pFDR < 0.05) between connectivity and tic reduction (termed R-map), signifying an optimal connectivity pattern. Lastly (4), the weighted connectivity of this R-map, limited to cortical voxels, to each other brain voxel was computed. This resulted in a map where each voxel contained the likelihood of matching the optimal connectivity profile, suggesting potential novel cortical target networks (termed target heat map).

Fig. 2. Location of electrodes and stimulation sites with average map of connectivity.

a Reconstruction of electrodes in standard space is shown in the upper and middle figures (the thalamus is highlighted in purple). b The distribution of stimulation sites is displayed (MNI coordinate: z = − 2), with most sites situated at the intersection of the ventral lateral (VL) and centromedian nucleus (CM), further encompassing the ventral anterior nucleus (VA) and the parafascicular complex (PF). The colour bar indicates the percentage of all stimulation sites per hemisphere (N = 37) covering the respective voxel. c on average, stimulation sites showed heightened connectivity to prefrontal, insular and parietal areas that closely aligned with the cingulo-opercular network (CON), further engaging the medial frontal and cingulate cortex. Within the primary motor cortex (M1), the inter-effector regions of the somato-cognitive-action network (SCAN) were positively connected, while the rest of M1, i.e., the effector regions, were anti-correlated. a = anterior; p = posterior; l = left; r = right.

Fig. 3. Association of DBS connectivity and tic reduction.

a In a region-of-interest (ROI) analysis, we correlated the percentage tic reduction after DBS with the respective connectivity of stimulation sites (N = 37) with the cingulo-opercular (CON) (top left) and somato-cognitive action network (SCAN) (top right). Non-parametric permutation testing using Spearman rank correlations (two-tailed, no correction for multiple comparisons) revealed a significant positive relationship between tic reduction and connectivity with the CON (r = 0.62, p = 0.0002) and SCAN (r = 0.47; p = 0.002), indicating that greater connectivity between the DBS sites and these networks is associated with more substantial reductions in tic severity. b Compared to other large-scale resting-state, only CON and SCAN showed a significant positive association with tic reduction (permutation test for two-tailed Spearman rank correlations, corrected for multiple comparisons with p_FDR < 0.05; marked with asterisks). c A post-hoc comparison revealed strong and significant associations between tic reduction after DBS and connectivity with the inferior (r = 0.596; p = 0.0002) and middle (r = 0.513; p = 0.0002) inter-effector regions, while connectivity to the superior inter-effector region showed a weak association (r = 0.276; p = 0.052) (permutation tests for two-tailed spearman rank correlations, not corrected for multiple comparisons). d Computing a data-driven, whole-brain voxel-wise model of optimal brain connectivity for tic reduction, termed R-map, clusters with a significant (p_FDR < 0.05) positive association between connectivity and tic reduction (red clusters) closely aligned with the CON. Within M1, the middle and inferior inter-effector regions of the SCAN were involved. Significant negative associations between connectivity and tic reduction are indicated in blue. e Subcortically, the insular clusters of the Tic-Reduction R-map further encompassed the posterior claustrum and small parts of the posterior putamen (z = 1). All scatter plots include linear trend lines with 95% confidence bounds. a = anterior; p = posterior; l = left; r = right.

Fig. 4. Target heat map for tic reduction.

a We computed a target heat map containing the voxel-wise connectivity to the R-map to identify a network profile that matches the optimal connectivity profile derived from thalamic stimulation (N = 37). The highest matches were observed within the cingulo-opercular network (CON). Within the motor cortex, the somato-cognitive action network (SCAN) showed the highest similarity of connectivity with the tic-reduction R-map. b Within the thalamus, the heat map peaked in the superior and anterior part of the centromedian nucleus (outlined in white). Connectivity of the cingulo-opercular network and the somato-cognitive action network also showed the highest connectivity in the centromedian nucleus, highly overlapping with the peak connectivity of the target heat map (all three maps are thresholded at r ≥ 0.2; slices taken at z = 1 and y = 20). a = anterior; p = posterior; s = superior; i = inferior.

Fig. 5. Replication in independent data.

a In an independent dataset of precomputed DBS sites (N = 10) obtained from the GTS-DBS-Registry, tic reduction was also significantly and positively correlated with functional connectivity (FC) to the cingulo-opercular (CON) (r = 0.054; p = 0.047) and somato-cognitive action network (SCAN) (r = 0.83; p = 0.003) (permutation-based two-tailed spearman rank correlation, not corrected for multiple comparisons). b Similarly, CON and SCAN became apparent in the heat map derived from this sample. c As a further replication, we investigated a previously published lesion network map. In the unthresholded lesion network map, derived from N = 22 lesions that induced tics, CON and SCAN were traceable as networks to which most lesions were connected. d The box plot displays the functional connectivity of each tic-inducing lesion to each large-scale resting-state network. Here, CON and SCAN showed the highest connectivity with respective lesions. Friedman’s test with post-hoc pairwise comparisons revealed that CON had significantly higher connectivity than all other networks except SCAN (two-tailed tests, adjusted for multiple comparisons with pFDR < 0.05). Asterisks indicate significant differences in connectivity. All results were highly similar for both hemispheres; only left-hemispherical results are shown for clarity. All scatter plots include linear trend lines with 95% confidence bounds. Box-plot elements include the centre line representing the median, box limits showing the upper and lower quartiles, whiskers extending to 1.5 times the interquartile range, and individual points. a = anterior; p = posteriorFigure 6: Peak Clusters of Target Heat Map. The top five clusters of the target heat map for tic reduction are shown (yellow, as derived when thresholding voxels at r > 0.6). The clusters peaked in the bilateral insula/operculum (left image, z = 3), the right supplementary motor area (SMA) (middle image, x = 6), and the bilateral supramarginal gyrus (right image, z = 27). All clusters resided within the cingulo-opercular network, shown in purple. The bottom of the figure shows critical evidence supporting the relevance of these areas for tic disorders in general and as neuromodulation targets^–. OLT = Open-label trial; RCT = randomized clinical trial; tDCS = transcranial direct current stimulation; fMRI = functional magnetic resonance imaging; rTMS = repetitive transcranial magnetic stimulation; a = anterior; p = posterior; l = left; r = right; sign. = significant.

Fig. 6. Peak clusters of target heat map.

The top five clusters of the target heat map for tic reduction are shown (yellow, as derived when thresholding voxels at r > 0.6). The clusters peaked in the bilateral insula/operculum (left image, z = 3), the right supplementary motor area (SMA) (middle image, x = 6), and the bilateral supramarginal gyrus (right image, z = 27). All clusters resided within the cingulo-opercular network, shown in purple. The bottom of the figure shows critical evidence supporting the relevance of these areas for tic disorders in general and as neuromodulation targets^–. OLT = Open-label trial; RCT = randomized clinical trial; tDCS = transcranial direct current stimulation; fMRI = functional magnetic resonance imaging; rTMS = repetitive transcranial magnetic stimulation; a = anterior; p = posterior; l = left; r = right; sign. = significant.